Frontier AI Safety Research Engineer

Data Science Jobs UK; Greater London; On-site; GBP 70,000 - 120,000; Full time

Job summary

The AI Security Institute, the world's leading team for understanding AI risks, is expanding its Control Red Team in the UK. We seek researchers and engineers who can run ML experiments, design adversarial tests, and deliver rigorous analyses.

You’ll work across frontier model evaluation, safety claims, and tooling, with access to large-scale compute and infrastructure support. We welcome strong scientists and engineers, with flexibility on seniority.

Qualifications

• Demonstrated ability to design, build and run ML experiments on frontier models with autonomous work on complex projects.
• Strong software engineering and ML experience: clean, reusable code for experiments including finetuning/inference frameworks.
• Understanding of how experiments support safety claims and how to critique them.

Responsibilities

• Design and run ML experiments on frontier models and attacks.
• Test frontier monitors and sandboxed environments; produce risk assessments for industry/government use.
• Build tooling/pipelines and use LLMs to automate evaluation loops.

Skills

ML research

Software engineering

Autonomous work

Team collaboration

Tools

Inspect

Finetuning frameworks

Inference frameworks

Job description

The AI Security Institute, the world's leading team for understanding AI risks, is expanding its Control Red Team in the UK. We seek researchers and engineers who can run ML experiments, design adversarial tests, and deliver rigorous analyses.

You’ll work across frontier model evaluation, safety claims, and tooling, with access to large-scale compute and infrastructure support. We welcome strong scientists and engineers, with flexibility on seniority.